Since, the packing of the genome within the nucleus informs cellular state, nuclear morphology and chromatin organization(NMCO) features hold biologically meaningful information. High resolution images of DNA as visualised using a fluorescent microscope is a convenient tool to characterize such DNA organization.
This package aims to provide an exhaustive set of interpretable morphometric and texture features for every single nucleus following segmentation from a 2D single channel image. Below is a brief overview of the features used
Documentation is being updated and will be made available here
The list of features and their description can be found in the file "chrometric_feature_description.csv"
Illustration of feature extraction can also be checked here
The current implementation has been developed in Python 3.
In order to avoid any changes to the local packages, install in a virtual environment (optional).
$ conda create --name nmco python
$ conda activate nmco
To clone the repository run the following from the terminal.
$ git clone https://github.com/GVS-Lab/chrometrics.git
Then install requirements and run the setup from the repository directory
$ pip install -r requirements.txt
$ python setup.py install
#import libraries
import os
from nmco.utils.run_nuclear_feature_extraction import run_nuclear_chromatin_feat_ext
# initialising paths
labelled_image_path = os.path.join(os.path.dirname(os.getcwd()),'example_data/nuc_labels.tif')
raw_image_path = os.path.join(os.path.dirname(os.getcwd()),'example_data/raw_image.tif')
feature_path = os.path.join(os.path.dirname(os.getcwd()),'example_data/')
# For a quick extraction of all available features for all labelled nuclei given a segmented image with default parameters
features = run_nuclear_chromatin_feat_ext(raw_image_path,labelled_image_path,feature_path)
Alternatively, use the CLI version using default parameters as follows.
python measure_nmco_features.py --rawdir <path/to/image> --datadir <path/to/labelled_image> --savedir <path/to/output/folder>
@article{venkatachalapathy2020multivariate,
title={Multivariate analysis reveals activation-primed fibroblast geometric states in engineered 3D tumor microenvironments},
author={Venkatachalapathy, Saradha and Jokhun, Doorgesh Sharma and Shivashankar, GV},
journal={Molecular biology of the cell},
volume={31},
number={8},
pages={803--812},
year={2020},
publisher={Am Soc Cell Biol}
}